SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size

24 Feb 2016  Â·  Forrest N. Iandola, Song Han, Matthew W. Moskewicz, Khalid Ashraf, William J. Dally, Kurt Keutzer ·

Recent research on deep neural networks has focused primarily on improving accuracy. For a given accuracy level, it is typically possible to identify multiple DNN architectures that achieve that accuracy level. With equivalent accuracy, smaller DNN architectures offer at least three advantages: (1) Smaller DNNs require less communication across servers during distributed training. (2) Smaller DNNs require less bandwidth to export a new model from the cloud to an autonomous car. (3) Smaller DNNs are more feasible to deploy on FPGAs and other hardware with limited memory. To provide all of these advantages, we propose a small DNN architecture called SqueezeNet. SqueezeNet achieves AlexNet-level accuracy on ImageNet with 50x fewer parameters. Additionally, with model compression techniques we are able to compress SqueezeNet to less than 0.5MB (510x smaller than AlexNet). The SqueezeNet architecture is available for download here: https://github.com/DeepScale/SqueezeNet

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Network Pruning ImageNet SqueezeNet (6-bit Deep Compression) Accuracy 57.5% # 14
MParams 1.24 # 4
Image Classification ImageNet-9 SqueezeNet + Simple Bypass Top 1 Accuracy 60.4% # 1
Image Classification ImageNet-P SqueezeNet + Simple Bypass Top 5 Accuracy 82.5% # 1

Methods